def run():
import argparse
parser = argparse.ArgumentParser()
parser.add_argument('--word_vec_path', default='data/crawl-300d-2M.vec')
parser.add_argument('--mode', default='train', choices=['train', 'eval'])
parser.add_argument('--datadir',
default='data/',
help='directory of input data')
parser.add_argument('--dataname', default='input.pkl', help='name of data')
parser.add_argument('--modeldir',
default='model',
help='directory of model')
parser.add_argument('--modelname', default='args', help='name of model')
parser.add_argument('--discourserank',
default=True,
help='flag of discourserank')
parser.add_argument('--damp',
default=0.9,
type=float,
help='damping factor of discourserank')
parser.add_argument('--epochs', default=10, type=int, help='epochs')
parser.add_argument('--batch_size', default=8, type=int, help='batch size')
parser.add_argument('--log_period',
default=500,
type=int,
help='valid period')
parser.add_argument('--opt', default='Adagrad', help='optimizer')
parser.add_argument('--lr', default=0.1, type=float, help='learning rate')
parser.add_argument('--norm', default=1e-4, type=float, help='norm')
parser.add_argument('--grad_clip',
default=10.0,
type=float,
help='grad_clip')
parser.add_argument('--keep_prob',
default=0.95,
type=float,
help='keep_prob')
parser.add_argument('--beam_width',
default=10,
type=int,
help='beam_width')
parser.add_argument('--length_penalty_weight',
default=0.0,
type=float,
help='length_penalty_weight')
parser.add_argument('--dim_hidden',
default=256,
type=int,
help='hidden dimensions')
parser.add_argument('--dim_str',
default=128,
type=int,
help='dimension str')
parser.add_argument('--dim_sent', default=384, type=int, help='dim_sent')
# for evaluation
parser.add_argument('--refdir', default='ref', help='refdir')
parser.add_argument('--outdir', default='out', help='outdir')
config = AttrDict(vars(parser.parse_args()))
print('loading data...')
train_batches, dev_batches, test_batches, embedding_matrix, vocab, word_to_id = load_data(
config)
config.PAD_IDX = word_to_id[PAD]
config.UNK_IDX = word_to_id[UNK]
config.BOS_IDX = word_to_id[BOS]
config.EOS_IDX = word_to_id[EOS]
n_embed, d_embed = embedding_matrix.shape
config.n_embed = n_embed
config.d_embed = d_embed
maximum_iterations = max([
max([d._max_sent_len(None) for d in batch])
for ct, batch in dev_batches
])
config.maximum_iterations = maximum_iterations
print('max iter:', maximum_iterations)
if config.mode == 'train':
train(config, train_batches, dev_batches, test_batches,
embedding_matrix, vocab)
elif config.mode == 'eval':
evaluate(config, test_batches, vocab)
